247zeroBETA2 : Motion and Extrusion - The Fastest Printer? (No, but it's cool anyways!)
ฝัง
- เผยแพร่เมื่อ 22 พ.ค. 2024
- To try everything Brilliant has to offer-free-for a full 30 days, visit brilliant.org/247printing/
The first 200 of you will get 20% off Brilliant’s annual premium subscription.
Let's see how I made the world's fastest Voron or maybe even the fastest FDM 3D printer!
Based on a Voron Zero ( Voron 0.2 ), I made a modded printer which printed the former world's fastest 3DBenchy in 02min25sec. It's called 247zero!
In part one of the mini series we check out what was necessary for the motion and extrusion system to do this! My 2min25sec 3DBenchy (247zeroB2) • "3DBenchy" printed in ...
You can build that printer 247zeroB2 NOW:
• 247zeroBETA2 File Rele...
247zeroBETA2/B2 modifications: tba
It is recommended to subscribe ;-)
Eddie The Engineer - GREAT and DEEP TH-cam channel:
/ @eddietheengineer
247printing Carbon Fiber X-Beam*: s.click.aliexpress.com/e/_DFp...
I used a Voron 0.2 basis for my mods - recommended sources:
-----------------------------------------------------------
Premium: LDO Voron 0.2 (3DJake) tidd.ly/41umJac
Good budget: Formbot Voron 0.2 (AliExpress) bit.ly/39kfWdC
Recommmended other FAST PRINTERS*:
----------------------------------------------
Voron 2.4 (AliExpress): bit.ly/3PklmWE
Voron Trident (AliExpress): bit.ly/3yxMaNe
SELECTION of the BOM - full BOM coming soon:
247zeroB2 Extrusion*:
--------------------------------------
Voron M4 extruder vorondesign.com/voron_m4
LDO 42STH48 2504AC 48MM Nema17 (AliExpress) s.click.aliexpress.com/e/_DdX...
Mosquito Magnum (5% off!) www.sliceengineering.com/247p...
Volcomosq by Nitram • How did I end up with ...
CHT Volcano nozzle 0.4mm (3DJake) tidd.ly/3VNkxZT
247zeroB2 Motion*:
--------------------------------
motor: LDO-35STH48-1684AH - 2.4A/58.5V (actively cooled) DON'T DO THIS!
Voron0.0 z-drive github.com/VoronDesign/Voron-...
Ceramic bearings s.click.aliexpress.com/e/_DkP...
Titanium fasteners s.click.aliexpress.com/e/_DcQ...
CPC rail for X fermio.xyz/chieftek-precision...
Lubrication (PTFE) amzn.to/2ZTE2rc
247zeroB2 electronics*
--------------------------------------
Only works with V0.0 z-drive:
BigTreeTech Octopus Pro s.click.aliexpress.com/e/_DEd...
4x BigTreeTech TMC5160pro s.click.aliexpress.com/e/_DkY...
or (works with all z-drives)
BigTreeTech Manta E3EZ s.click.aliexpress.com/e/_Ddb...
4x BigTreeTech EZ5160(pro) s.click.aliexpress.com/e/_Ddb...
BigTreeTech TFT35 SPI V2.1 (screen) s.click.aliexpress.com/e/_DF0...
.
.
and more...
Additional Information*:
----------------------------------------
My 3min27sec 3DBenchy (247zeroB1) • 3DBenchy printed in 03...
Beraval's 02min40sec Benchy • 2min40sec Benchy on a ...
Stefan's Flow-Test-Tool: • Easy Hotend Benchmark ...
0:00 Intro: What's the point on this?
1:57 Extrusion System
6:14 Sponsor: Brilliant
7:18 Motion System
13:23 Conclusion and Outlook to Part 2
This video was sponsored by Brilliant.
My modest video gear
-------------------------------------------------------
iPhone 11 Pro: amzn.to/2VZYswJ
Smartphone video rig: amzn.to/3iriub5
GoPro Hero 9: amzn.to/2UZI9Q5
FLIR ONE PRO: amzn.to/2UWAe67
Similar to the soft boxes I use: amzn.to/3xRes2d
Similar to the tripod I use: amzn.to/2V1I6mw
*Can be affiliate links at no extra cost for the customer. - วิทยาศาสตร์และเทคโนโลยี
![[1595] Sentry Safe’s Lock Design Malpractice](http://i.ytimg.com/vi/aZUDFrqOSvM/mqdefault.jpg)
![[1595] Sentry Safe’s Lock Design Malpractice](/img/tr.png)
To try everything Brilliant has to offer-free-for a full 30 days, visit brilliant.org/247printing/
The first 200 of you will get 20% off Brilliant’s annual premium subscription.
Sorry for the re-upload - Adobe Premiere TRICKED ME :-)
And here I thought having only sound was an interesting choice
Sounds like something Adobe software would do 😠
Happens a LOT unfortunately. They can always relate on the hardware, drivers and whatever used, but my system is really quite common..
LOL. 247radio !
That is okay, you got a new subscription out of it.
acceleration should be mesured in G at this point
This printer is mad... hop the next videos are at the same level as this master piece...
Even faster then your delta. But your delta look more sick when it prints!
Thanks a lot Nitram - did you just write "MASTER PIECE" ?!
This is all absolute madness and I am here for every bit of it! Thanks for sharing as always!
🥰
I love the absolutely insane printer movements showcased in the real-time demos! Thank-you for sharing all the work you've put into this!
Thank you, I am glad you liked it!
nice one Albert !! that printer is a beast !!
Awesome build! See you in part 2 ;))
Thanks Jan - see ya!
I’m looking forward to seeing how much torque and speed output we can get from NEMA14 motors! I have a hunch the rated current is heat limited and not due to saturation, so we may have a lot to play with yet 😄 🔥
Nema 14s tend to have higher inductance, especially for the torque you get out of them, so the limiting factor will be back EMF. The ones I've seen are rated at 3.8Mh for 40Nm, that's gonna kill your max speed, even at 48v, that might hit 1000mm/sec, not much more, even running them under max current. Compare that to LDO nema 17s - much lower inductance at 1.5, and you can under current them and still get more torque and at 48v you can push over 2400 mm/sec.
cant wait for the details! :)
Could you in the future also explore the longevity of the motion system (belts, pulleys, bearings, etc) running at such speeds please?
On the list!
Like always crazy. Geil die outtakes.
Ich mag jeden, der die Outtakes mag - der Abgrund rein in meine schwarze Seele!
Love ur videos man, such great work…keep it up, it’s greatly appreciated.
thats so awesome
next you should make the printbed move to increase the relative speed between it and the nozzle
13:48 The fly did his inspection and he approves.
Haha, you are a great observer! Just realized it when re-watching after uploading :-D
Thanks for sharing!
How does the hotend move that fast without flying off?! Next Level stuff mate! Just f'ing amazing!
Thanks, man - low weight helps a lot here!
can't wait for part2
Keep up the great work loving the videos
Increasing voltage makes sense, if you've decreased the mass you're moving, then the goal would be to run at the lowest current setting you can use to generate the torque needed. Coming from cnc world, the reason for using 60v PSUs is to get decent speeds at high torque as the goal is to overcome emf. Using low inductance steppers will get the best results. You can increase current while not hitting max rpm due to emf, there are emf calculators that can help you find the max theoratical speed you can achieve. Try to lower your current while running at 50v, and you should be able to hit higher speeds. With a light gantry and extruder, you're not torque limited.
Thanks a lot for that very precious comment - didn't know that 60V is common in the CNC world! As a FDM printer is also a CNC - why not go at high voltage...?
@@247printing I run my 5'x3' CNC with hybrid servos on 60V DC because if you do the math EMF on 48V would limit my travels. CNC world, no one is even trying to go over 200mm/sec, it's not practical to move 30-50kgb gantry+spindle faster. Keep accels to 200-300mm/sec -- yea, very different than a 3D printer. Main thing to keep in mind is current is torque, voltage is speed. CNCs need a LOT of torque, there is the moving weight, and there is the resistance of what you're cutting, wood, metal etc. So my Y axis needs a nema 34 at 6A of current(max). When you calculate the back EMF at that current, you need more voltage. I even tried running it at 90v DC, but in US with 120 AC mains, a 90V DC PSU is really sketchy. If 60V DC setup wasn't going to work, I was going to move to 240 AC, since my spindle runs on 240 already. For 3D printers you can absolutely go up in voltage, but you need external drivers that can support it. Step sticks and other small trinamics, 55v as you're doing is really pushing it. There are plenty of really good external drivers that can run on 60-90V DC. In Europe with 240 from mains, you can get a 90V DC PSU to run very nicely. Look up EMF max RPM calculators, plug in your current, inductance and voltage -- boom you get theoretical max motor RPM, with corexy it will take a bit of math to calculate RPM relation to actual speed. I have a big printer with 2 heavy gantries that I've lightened as much as I can so far, and I run them on 2 nema 23s on each side and I lower the current to under 50% of max rating because the nema 23s provide much more torque than I need, but I can push speeds higher, for a big printer I can hit 500m/sec at 10k acceleration and only at 24v. If I max out the current 300 mm/sec is max. Printers use belts, usually, so 1 RPM = even with a 16t pulley is 32mm of movement, on a CNC, like mine with direct driven ball screws, 1 RPM is 5mm of movement, you need a lot more RPMs to hit target speeds, thats why on bigger CNCs, you either go with higher pitched ballscrews (I wanted precision, so I didn't) or you go with servos and more voltage so you can spin them faster.
This... Is... Awesome! 😁
An extremely informative video.
What do you work on aside from this type of Content?
I'm new to 3D printers in a sense - but I've spent a lot of time on design of military aircraft - they go fast in a different way.
I really appreciated the blend of discussion of the electrical and mechanical elements - having a bit of a unique education across electrical, mech, and material science I find this fascinating.
Might be time to break the rules and step up to servos! You could make the gantry stiffer again for less overshoot, more quality
Aren't those >=Nema23? Absolute overkill for that tiny printer :-D I don't like the idea really...
@@247printing Hi Albert,
My reply must have gotten lost somewhere, there are DC servos in nema 17 size available! Check the JMC iHSV42 servos, they cost ~90€ and are available as 58 and 72 watt versions, and are fully rated with full torque up to 4000 rpm. I used the nema23 version in a high temperature printer i built for the Solvay AM cup. It moves my 10kg heavy Y axis easily at >5g‘s with no lost steps in sight, it really is fabulous, no regrets, all while being completely dead silent. Don’t hesitate to ask me questions about them if you want.
Grüße
Max
Really game changer beast bro
@3:04 Planning on tuning that 500 as well? Just ordered a Goliath as well, with the WC Hextrudort. Mine will never be a proper speedprinter, but this hotend/extruder combo seems amazing for speedprinting technical materials in enclosed printers.
I have a kit here, yes! I want to do some modest speeding up while staying at quality - need the size for a project!
Once the core of the motors saturate, the have the same magnetic permeability as air. That's why your acceleration still increases after you saturate them, but at a lower rate. This is because the magnetic permeability is much higher in unsaturated ferrite core. Also, torque is proportional to magnetic field, which is somewhat proportional to magnetic permeability of the core.
Thanks a lot, I'll check the theory on those key words!
What i have found out: using stiffer filament for the xy joints really helps with acceleration. On my v0.2 with a fat boi lgx lite, microprobe, canbus board and bigger 4010 fans on the toolhead i got additional 2,5k acceleration by just switching from regular abs to glass fiber reinforced abs. Recommended acceleration went up to 16,8k from just above 15k, even though the glass fibers increased the moving mass ever so slightly. So you may want to try a stiffer filament for some quality improvements. Here i could recommend pc gf or pa cf from cr3d or the super strong filament from igus
Light weight and strong/stiff does the trick yeah! Though your rather "slight" improvement (+12%) could also be due to belt re-tensioning in between the change? Just guessing!
@@247printing could be, although klipper never has recommended in that region before the switch. Did several tension settings, all about the same.
Oh, forgot to mention i also printed the x carriage out of the abs gf20, i can imagine that this part has an even bigger influence than the xy joints
I've always wanted to try out C02 cooling. Feed liquid CO2 directly to the head, allowing sublimation of the CO2 to cool the part super rapidly. What are your thoughts?
I think this would have to be tried out! No clue (yet) how to, but sounds really interesting!
8:51 i don't have a tick tock
You don't really need - TH-cam is sufficient :-)
If it's fun, then it's not useless. Love your stuff !
Also I remembered when you said you used this big Z motor for fast layer change : why not using a belted Z instead ? Could've made the BOM lower and the back of the printer less crowded ?
Thanks mate! That's also one of my philosophies!
I’m really curious if an airjet solid state fan moves enough air, decreases enough mass, and/or decreases enough vibrations to make any sort of useful difference.
Damn, amazing!
Awesome stuff, just learning about it all now.
I'm guessing you've thought of all this before but just asking for learning purposes:
Would a stiffer frame not help at all? HBR25 rails or whatever CNC machines use?
What about a more sophisticated base / platform to minimize vibrations?
Maybe even some kind of active vibration compensation system?
Will geared steppers work? Or bigger pulleys. Might lose some resolution though.
Are the cooling shrouds for the NEMA 14s available somewhere? I think they where already part of the V0.0 version but I cannot find it anywhere...
Beware of an "invisible" problem there. The umbilical breakout board. AFter lots of troubleshooting I found out my umbilical cable connectors were not as solid as I thought, especially the backside one which connects to the fixed oblong board. The issue is the continuous back-and-forth movement of the head, aggravated by the relatively short umbilical cable lenght, causing the conector(s) to undergo rather extreme tilting angles. That may cause brief interruptions in some of the cables. This problem is equivalent to introducing a randomized PWM in some of the cables connected to the print-head, generating all kinds of nasty consequences like faulty heaters, false temperature runway fails, lagging thermistor data and extruder stepper with unstable voltage/current supply. It is challenging to diagnose this problem in real-time with standard metrology equipment like cheap multimeters but you may be able to see it with the printer turned off, measuring the resistance of each connection while gently applying back and forth pressure to the each end of the umbilical cable. I just got rid of mine after all.
Have you considered using carbon fiber for all of the moving parts? You can get crazy stiffness and lightness. It's also a material that benefits greatly from being able to know the direction of force before manufacturing which is exactly what's needed here. Easy Composites has a bunch of tutorials on TH-cam explaining different process to make carbon fiber. They also sell carbon fiber rods and are UK-based.
13:48 HouseFLY!!!!! Lol
😁
13:13 if there is a difference between theory and what is measured then the theory might be missing something.
Unless the rotor was taken out and modeled I'd say it probably wasn't fully saturated if you got more torque from it.
13:26 YES i have 1 why you have to use the cooling block on the hot end you could use a airgap - because this printer is fast and gets extra air from the fans.
2 It would be interesting to use a lineal motor on the Y axis , like the motors in pick and place machines.
ok you could say ok OP why dont you build that , the same reason I have no time , and place mostly space.
I'm actually working on a printer with a linear magnetic motor motion system but it's slow going, magnets are expensive and doing all the FEA simulations for the magnetic fields and forces has been a whole lot more work than I anticipated.
@@oliverer3 Yes that is the name of the game, after a point the work or money is exponentional
The poop should be a benchmark for how much a printer can extrude within a specific period for a given z distance.
Moin! One question regarding going even faster in the future. Do you think systems, like the Annex-Engineering K3 with using only the linear rails for X and Y and 4 motor drive, will be the future? I think it would probably be a bit less stiffer than the V0 but if you someone would build one with a smaller print volume, for more stiffness, and a bowden extruder, it would be a propper challenger for the V0 design.
That mechanical system is pretty much the Ultimaker style of cartesian with motors on frame, but with extra motors and linear rails instead of the circular bars.
One advantage is a symmetrical weight distribution on X and Y.
Servus! Yes I am very tempted by that motion system (CroXY) and I have a feeling about it being the next big thing in the near future.
(What CoreXY just became in the mainstream market)
@247printing maybe you could use LWPLA for the structure parts. Use normal pla for the shell and stiffness but use the LWPLA for the infill and you could shave some weight
@Broskisnowski The thing is that the normal PLA has more strenght than the not foamed LWPLA at my institute we did test for this and the normla pla was stronger than the LWPLA (not foamed)
I wonder how complicated it'd be to belt a stepper to the top of the toolhead to get all the benefits from a direct drive extruder without the weight of the motor. You'd probably have to make a spring tension system because it moves around, but it'd help quality. Might even speed it up because better control over pressure.
Insane printer! I’m wondering if you have a name/any resources for the cooling solution you use with the fan ducts attached to the frame. I’d like to use something similar, but cannot find any information whatsoever about this style of design. Thanks
Thx.! These are my designs and not released yet. Coming soon as soon as I am happy with the quality.
In your opinion what’s the most important thing to keep in mind for a fast corexy (motors, speed, weight…)
Clever segway :)
Any tips for high temp PLA, which could be used as a substitute of Uber-PLA?
Do you model those cross beams? I have a license solidworks simulation if you ever wanted to FEA them before manufacturing
Some violent vibrations! Concrete frame might help?
Is it possible to scale the printer and have the same type of speed? Build volume of 1000mm x 1000mm x 1???mm ? Also, what is the quality of parts printed at these high speeds and the durability? Love your videos by the way, subscribed to you channel and newsletter.
More phases for higher torque in the stepper motors might help right?
What filament is that in the box being used ?
Watching the extruder extrude those massive piles of filament….. I could smell it through the screen, the smell of my printer has been ingrained into my brain
very cool!
do you know MFI value of that filaments? I'd like to see comparison standard MFI 8 and high flow 30-60g/10min.
Are those XL grippers compatible with the normal 0.2?
For motor cooling at high current, have you considered something beefier than those 4010 noctuas? They produce neither high airflow nor high static pressure, so they are not ideal for this kind of application. Larger fans (4020, 4028, 6025) or liquid cooling (like Vez is doing on his steppers) could let you achieve lower operational temperature.
Very good point! I had a line in the script like "For those currents I'd need to step up the fans for the stepper cooling", but I deleted it, because I don't want to encourage the public to try those high temps, before I did more testing.
@@247printing a simple water cooling setup would work nicely and overcome the surface area/local space limitations. There are even pre-made water blocks for stepper motors, although I’m sure the quality varies quite a lot and a trio of completely closed-loop CPU coolers May actually be easier.
What do you think about a odrive to run the core xy with Servo motors?
I want the files for them feet on the voron, they look good on there and now im going on a mission to find them or make them
Great wedeo
I think it's time to start breaking the speedboat rules. We can go even faster!
For sure, but for comparability it's nice to stick to it.
Prusa showed with the MK4 gcode how much faster it can be with lowering walls etc.
@@247printing Sure, walls are one thing. I'm thinking less in the way of changing the slice, but more like moving to BLDCs for faster motion. A second look at the constraints would be neat.
considering the benchy was recognizable as one ... huge success with much room for improvement
Albert, I got a question for you. I also got to tell you I stopped working my extruder I told you about. I was trying to solve a problem that didnt exist...Anyway:
Could you link me to the printed X gantry you used on the old 247 zero? You said you printed it out of PET-CF. I want to print that as a start for my 247 zero build, just because I dont want the price to shoot up immediately.
I half expect to see some kind of liquid cooling for the steppers to show up at some point. Just to see if one can squeak just that much more from them, I admit I have no clue how one would do it as I personally have doubts on those water blocks I have seen that just screw onto the back plate. Seems like one would need to get the sides of the motor as well.
The benchy is ugly but this is NHRA Top Fuel Printing, The whole idea is brute speed. after all you do not play at 11 for audio fidelity you do it because its one louder.
When you're talking about the rated current of the motor, 1.68A in your example. It only means that it is the continuos current where the motor can operate without getting too hot. This has been set by the manuafacter and it does not count active cooling. So you can exceed the rated current if you are able to keep the motors temperature low enough for it to not break.
So in that sense exceeding the rated current is not a limit of your motors theoretical capabilities. The higher current you give to your motor the more torque it has. But when you increase the frequency of your pulses there comes in another factor that is inductance. Inductance (milliHenries, mH) tells how fast your coil is able to reach its maxium current. Let's say your motors coil achieves its fully energized state in 5ms. If your steps take 9ms each, the coil has time to be fully "energized" for 4ms. It is already less than half of the step at the maxium torque. When you increase the stepping frequency even higher for example 3ms, your motors coil has no longer time to be fully "energized". In the following example the slow rising of the current caused by the inductance is linear.
We can only reach 3/5 (calculated 3ms/5ms) of the maxium current each step. In this situation increasing the current helps because 3/5 of 3 amps is a lot more than of a 2 amps.
In summary choose a low inductance motor, use as high voltage as possible and make sure the current is as close to maximium as possible without destroying the motor. There's a good and short post about this if you google search "LinuxCNC Stepper motor inductance".
Thank you SO MUCH!
I love this dudes wideos 😅
I love making Phüdijos
Bro is over here printing in the year 3000
Schönen Abend. Frage: Wann wird Version A2/B2 als Model bereit sein?
This might be something extremely dumb, but have you considered stepping down the grade of belt for the x/y. I am not sure what belt size is on there currently but it would be like going from HTD5mm to GT3mm(those are some belt sizes I have used). It could save a few grams of the moving mass and potently reduce the size of the bearings. The big down side to this is it would increase the amount of stretch in the belts.
Very nice idea - didn't even know there are small ones. Maybe those are sufficient!
This might be an ignorant question, but... why don't you just use the pathing of a 3D Benchy for testing? Scale it up if you want the printhead to move a lot more than a 100% benchy, but otherwise it has all the properties of what the printer does during a print.
Sorry that i even have to ask, but wHAt filLamEnT Did yoU USe?
What kind of filament is that
Bolt it to the wall. When you come down to THAT speed, the whole body, and the table it self, just shakes.. Even better if u get a corner, and anchor 2 sides.
nice, i see your next project should be watercooling your motors :D
Have you tried sorbothane to greatly reduce vibrations ? This is magic material to dampen
Accelerometers and an algorithm compensating for movements could greatly improve print quality.
Do you see AWD feasible? There seems to be room at the front idlers :)
Smooth sponsor transition there lol
😇
mi piacerebbe avere i piedi della stampante com ei tuoi, cosa devo fare per averli? grazie
where can you find the stl files for the widened feet?
Not published yet
I would love to test the circle speed file 😅
" i might be a German engineer, but considering I'm really behind on my schedule" sounds like you might be an English engineer
Do you plan on switching to BLDC motors ? They are faster and have higher efficiency.
would not comply with the speed benchy rules
Wow! Next, bolt it to a concrete floor and use input shaping.
Faint Vez3D laughing hysterically.
Have you considered some vibration dampening? For example if you extended your vertical frame into a block of gravel/epoxy that might dampen the vibration a little bit and give you better quality.
can someone link me the macro to test acceleration shown on the video?
He wrote it, it's just a string of g1 commands
Why don't people like racing clean benchies? Seems like just as interesting of a challenge and much more useful?
I wonder if you could bowden high speed cold air into the print
berd
Be nice to see the quality of your prints at a reasonable speed like 200/mms
Omg I wanna build this little monster so bad... mount a Goliath / HextrudORT on it
As it seems part cooling is the most important part that is slow.... how about using evaporation to achive extremely fast cooling? Water evaporation is extremely fast and cant be beaten with air alone. A small piezo mist maker inside the head and micro tube (from a inkjet printer) that provides the water and on the backside of the printer a peristaltic pump that pushes very very little ammount of water to the piezo mister. This should give you extreme cooling ability's with the fans you already have.
What do you think about this idea? Yes there are difficultys like preventing the mist getting on top of layers where you want to print, but it might work even with that because the melted plastic is at at least 200C which means the water evaporates immediately....
Would love to hear / read what you think about this....
Cooling is not an issue on this printer. Watch the full video :)
Still not STl files. I want to print the foots.
Bowden extrusion like "direct who?"
What I'd like to see is personal user trying to push resin printing to it's limits
Any news on this? :D
use ac servos
Would it be cheating to double-gang the NEMA14 in series? That would benefit from the voltage… 😅
Hmmm, what is double-gang?
@@247printing this is an idea but I have no experience of, so take with a grain of salt / skepticism. I’m a doctor, not an electrical engineer.
(I think) Double-ganging usually means running two switches in parallel, or two switches from a single line, but here I mean two stepper motors together, or two stepper drivers together.
Mechanically, mount a second stepper motor above the frame and mechanically link it to the other stepper motor.
If the stepper motor is magnetically saturated at that current from a single driver, then you might try having two stepper motors in series, at high voltage, from the same stepper driver. This is dependant on the internal wiring and if there is a common ground may not be possible; you may need to switch the order of windings if the motors at facing each other.
Alternatively, if the stepper motor is not magnetically saturated and the stepper driver is the limiting factor, driving the same motor with two drivers in parallel. This could be one stacked o. Top of the other from the same socket, but iirc some boards let you use multiple sockets (for example when you have two z-axis motors) configured in firmware. Send them both to the same motor.
I did wonder if the reason that torque maxed out at the lower voltage was that you were setting the current _limit_ on the driver, but at that lower voltage the resistance of the stepper motor windings meant that no more current could flow (I=V/R) and that the extra voltage meant the driver could increase the current.
This is me not asking about the secret PLA